Pregnancy history may contain clues about metabolic health years after reproduction has ended. A new prospective analysis of women in the United States suggests that a reproductive-history measure capturing a person’s propensity for miscarriage is associated with a higher subsequent risk of type 2 diabetes, even after researchers accounted for a corresponding history-based propensity for gestational diabetes.
The study, published in Scientific Reports on 3 October 2026, analysed data from the Sister Study, a large US cohort established to investigate health and environmental risk factors among women. The researchers did not simply classify participants according to whether they had ever experienced a miscarriage. Instead, they developed an empirical Bayes measure that incorporated the sequence of pregnancy outcomes together with factors that influence miscarriage risk, including maternal age and smoking.
A reproductive history rather than a single event
The analysis included 42,313 participants who had experienced at least one pregnancy. For pregnancy-level analyses, the researchers updated each woman’s estimated miscarriage propensity as her reproductive history unfolded. This allowed them to test whether earlier reproductive outcomes predicted what happened in later pregnancies.
A separate propensity measure was created for gestational diabetes mellitus, with body mass index incorporated into that estimate. This distinction matters because gestational diabetes is already recognised as a strong marker of later type 2 diabetes risk. The researchers therefore wanted to determine whether the information contained in miscarriage history contributed additional predictive information rather than merely reflecting a known gestational-diabetes pathway.
The pregnancy-level results showed that the miscarriage propensity measure strongly predicted subsequent miscarriage. For every 0.1-unit increase in prior miscarriage propensity, the odds of miscarriage in a subsequent pregnancy were 5.14 times higher, with a 95% confidence interval from 4.20 to 6.28. The same 0.1-unit increase was also associated with higher odds of subsequent gestational diabetes, with an odds ratio of 1.51 and a 95% confidence interval from 1.02 to 2.22.
Following women after reproduction
The central long-term analysis focused on 39,468 women who had completed reproduction before entering the Sister Study and who did not have a diabetes diagnosis at enrolment. The researchers then followed these participants prospectively for incident type 2 diabetes.
Across 532,364 person-years of follow-up, 2,874 participants developed type 2 diabetes. Cox proportional hazards models were used to estimate how the reproductive-history propensities related to later diabetes incidence while adjusting for relevant factors.
Each 0.1-unit increase in the history-based miscarriage propensity was associated with a 38% higher hazard of incident type 2 diabetes. The hazard ratio was 1.38, with a 95% confidence interval from 1.13 to 1.69. The gestational-diabetes propensity measure was independently associated with later type 2 diabetes as well, with a hazard ratio of 1.35 per 0.1-unit increase and a 95% confidence interval from 1.29 to 1.41.
Importantly, the association involving miscarriage propensity remained when gestational-diabetes propensity was included in the analysis. This suggests that a woman’s pattern of pregnancy losses may capture metabolic or physiological information that is not fully represented by her history of gestational diabetes.
What the association may mean
The findings do not establish that miscarriage causes type 2 diabetes. A more plausible interpretation is that pregnancy outcomes and later metabolic disease may partly reflect shared underlying susceptibility. Pregnancy places substantial demands on cardiovascular, endocrine and metabolic systems, meaning reproductive events can sometimes reveal vulnerabilities that become clinically apparent only years later.
Previous research has linked pregnancy loss with later cardiometabolic outcomes, but the new study takes a different analytical approach. Rather than treating miscarriage as a simple yes-or-no exposure, the researchers attempted to summarise the pattern of reproductive outcomes while accounting for age-related miscarriage risk and smoking. This can distinguish, for example, between a loss occurring at an age when miscarriage is relatively common and a reproductive history containing more losses than would otherwise be expected.
The association between miscarriage propensity and later gestational diabetes also supports the possibility of shared biology. Insulin resistance, vascular function, inflammation, hormonal regulation and other metabolic processes could plausibly influence both pregnancy maintenance and later diabetes risk. The present observational study, however, was not designed to determine which biological mechanisms explain the association.
Why the numbers need careful interpretation
The reported hazard ratio of 1.38 applies to each 0.1-unit increase in the researchers’ estimated miscarriage propensity. It should not be interpreted as meaning that every individual miscarriage raises a woman’s diabetes risk by 38%. The exposure is a model-derived measure based on reproductive history, not a count of miscarriages.
Similarly, a hazard ratio describes a relative difference in the rate at which new diagnoses occurred during follow-up. It does not mean that 38% of women with a miscarriage history will develop type 2 diabetes. In the prospective analysis, 2,874 cases occurred among 39,468 women over more than half a million person-years of observation.
The study also cannot show that changing miscarriage risk would necessarily change later diabetes risk. Unmeasured factors could contribute to both reproductive outcomes and metabolic disease, and reproductive histories were reconstructed from participant information rather than from a randomised intervention. The Sister Study population also has its own recruitment characteristics, including enrolment of women whose sisters had been diagnosed with breast cancer, which should be considered when generalising estimates to all populations.
A possible additional window into long-term metabolic risk
Gestational diabetes is already an important clinical signal for future type 2 diabetes. The new findings raise the possibility that a more complete reproductive history could add useful information to long-term metabolic risk assessment, particularly when pregnancy losses occur in patterns that are unusual after age and other known influences are considered.
That possibility requires further validation before a miscarriage-propensity score could be used clinically. Researchers would need to test the approach in more diverse populations, determine whether it improves prediction beyond routinely available risk factors, and establish whether additional screening based on reproductive history leads to better health outcomes.
For now, the study adds to evidence that reproductive history can function as more than a record of past pregnancies. It may also provide a window into physiological vulnerabilities that matter for health decades later. The key contribution is not that miscarriage should be treated as a direct cause of diabetes, but that patterns across pregnancy history may contain information relevant to future metabolic disease.
Source Information
Study: Venkat, V., Goldblatt, D., Rogers, M., Diaz Santana, M. V., Weinberg, C. R. et al. “History of miscarriage and type 2 diabetes.” Scientific Reports (2026).
Published: 3 October 2026.
DOI: 10.1038/s41598-026-71644-6.
Study type: Prospective cohort and pregnancy-history analysis using Sister Study data.
Funding: Intramural Research Program of the US National Institute of Environmental Health Sciences.
